Meta-Llama-3-8B-Instruct
24 records across 2 hardware.
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| MI50 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 310.24 | – | – | – | 2026-10-03 | community | – | source |
| MI50 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 318.84 | – | – | – | 2026-10-03 | community | – | source |
| MI50 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 362.25 | – | – | – | 2026-10-03 | community | – | source |
| MI50 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 426.41 | – | – | – | 2026-10-03 | community | – | source |
| MI50 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 169.92 | – | – | – | 2026-10-03 | community | – | source |
| MI50 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 406.99 | – | – | – | 2026-10-03 | community | – | source |
| MI50 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 181.72 | – | – | – | 2026-10-03 | community | – | source |
| MI50 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 454.36 | – | – | – | 2026-10-03 | community | – | source |
| MI50 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 197.86 | – | – | – | 2026-10-03 | community | – | source |
| MI50 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 531.68 | – | – | – | 2026-10-03 | community | – | source |
| MI50 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 204.92 | – | – | – | 2026-10-03 | community | – | source |
| MI50 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 571.23 | – | – | – | 2026-10-03 | community | – | source |
| gfx1152 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 69.09 | – | – | – | 2026-10-03 | community | – | source |
| gfx1152 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 92.57 | – | – | – | 2026-10-03 | community | – | source |
| gfx1152 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 98.26 | – | – | – | 2026-10-03 | community | – | source |
| gfx1152 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 117.49 | – | – | – | 2026-10-03 | community | – | source |
| gfx1152 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 128.82 | – | – | – | 2026-10-03 | community | – | source |
| gfx1152 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 128.20 | – | – | – | 2026-10-03 | community | – | source |
| gfx1152 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 135.69 | – | – | – | 2026-10-03 | community | – | source |
| gfx1152 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 135.13 | – | – | – | 2026-10-03 | community | – | source |
| gfx1152 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 191.79 | – | – | – | 2026-10-03 | community | – | source |
| gfx1152 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 188.99 | – | – | – | 2026-10-03 | community | – | source |
| gfx1152 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 204.13 | – | – | – | 2026-10-03 | community | – | source |
| gfx1152 | Meta-Llama-3-8B-Instruct | Q2_K | llama.cpp (ROCm) | – | 205.85 | – | – | – | 2026-10-03 | community | – | source |